Note: Descriptions are shown in the official language in which they were submitted.
SYSTEM AND METHOD FOR MEASURING AND EVALUATING BUILDING
ENERGY PERFORMANCE
TECHNICAL FIELD
The present invention relates to a system for measuring and evaluating
building
energy performance and method for driving same, more particularly, to perform
building
energy performance evaluation in view of a change of outdoor-air temperature
which has a
major influence on building energy consumption before and after an energy
retrofit of
building, based on the measurement data of building management server A
(building
automatic control system) or building management server B (energy monitoring
system).
BACKGROUND ART
The prior art describes a statistical method to search change point (or
turning point) of
building energy regression model according to outdoor-air temperature, in
which interval
(Grid) is set evenly by dividing difference between maximum and minimum values
of
outdoor-air temperature of population by a predetermined number (e.g. 10) and
outdoor-air
temperature at which a residual (root mean squared error, RMSE) of the
building energy
regression model for each interval is lowest is set as a change point.
However, when the difference between the maximum and minimum values of
outdoor-air temperature of population is large, the interval (Grid) of the
outdoor-air
temperature for searching the change point becomes large. As a result, there
are
disadvantages in that an accuracy of a searched change point and a reliability
of a regression
model are inferior.
1
CA 3005184 2019-08-16
Further, in the prior art, there are disadvantages in that the accuracy of the
searched
change point and the reliability of the regression model according to a
distribution of the
population are not always increased even if the interval (Grid) of the outdoor-
air temperature
is arbitrarily set to be narrow, and if the interval (Grid) of the outdoor-air
temperature for
searching the change point is set too narrow, then calculation speed is slowed
down.
DISCLOSURE
TECHNICAL PROBLEM
The present invention has been made to solve such a problem, and it is an
object of
the present invention to provide a system for measuring and evaluating
building energy
performance and method for driving same which conducts building energy
performance
evaluation considering a change in outdoor-air temperature that has a major
influence on
building energy consumption before and after energy retrofit of building using
building
management server.
Further, the present invention has been made to solve such a problem, and it
is other
object of the present invention to provide a system for measuring and
evaluating building
energy performance and method for driving same which conducts a statistical
analysis and a
reliability analysis by generating a regression model having a change point
according to
measured data.
Further, the present invention has been made to solve such a problem, and it
is an
another object of the present invention to provide a system for measuring and
evaluating
building energy performance and method for driving same which select virtual
change
points sequentially for outdoor-air temperature of population and set an
intersection point of
2
CA 3005184 2019-08-16
a left side regression model and a right side regression model using a least
square method
with each virtual change point as center in order to search for a change point
of a regression
model generated according to measured data.
TECHNICAL SOLUTION
In order to achieve the above objects, a system for measuring and evaluating
building
energy performance according to the present invention conducts measurement and
evaluation of performance regarding building energy using building management
servers A
and B and comprises: a measurement variable setting module for receiving
measurement
data, which includes a measurement history, from the building management
servers A and B
and setting building energy and an outdoor-air temperature as variables; a
measurement data
analysis module for receiving the variables, related to occasions before and
after energy
retrofit for saving the building energy, from the measurement variable setting
module and
analyzing a correlation between the building energy and first and second
outdoor-air
temperatures using an analysis graph showing trend of and correlation between
the building
energy and outdoor-air temperature before and after energy retrofit; a
regression model
generating module for plotting the variables related to the occasion before
energy retrofit,
which have been output from the measurement data analysis module, generating a
building
energy regression model with a change point of the first outdoor-air
temperature, and
applying the second outdoor-air temperature, related to the occasion after
energy retrofit, to
the building energy regression model, thereby generating a reference model; a
statistical
analysis module for statistically analyzing an energy retrofit model with
regard to the
building energy regression model and the reference model, which have been
output from the
3
Date Recue/Date Received 2020-10-20
regression model generating module; and an energy saving analysis module for
comparing a
difference between a predicted amount of energy consumption, which is one of
data
resulting from statistical analysis by the statistical analysis module, and an
actual amount of
energy consumption, which has been output from the building management servers
A and B,
thereby extracting an amount of energy saving.
The system for measuring and evaluating building energy performance registers
one
or more monitoring and control points corresponding to facility generating the
building
energy before analyzing the correlation, and then collects the measurement
data using any
one of a work tree type configuration method or an individual energy metering
installation
method for the facility.
The analysis graph according to the correlation uses a time series plot or an
X-Y
scatter plot.
The building energy regression model is at least any one of an average model,
a
simple regression model, a heating regression model, a cooling regression
model, a cooling
and heating regression model, or a general regression model, according to
shape of the
analysis graph and a number of the change points searched from the analysis
graph.
A virtual change point is sequentially selected from a minimum value to a
maximum
value of the measurement data and a factor in which sum of residual (Mean
Squared Error,
MSE) for left side data and right side data is minimum is searched using a
least square
method with each virtual change point as center.
Any one of a number of the measurement data, a number of the change points of
the
building energy regression model, coordinates of the change point of the
building energy
regression model, slope of the left side model or right side model of the
building energy
4
Date Recue/Date Received 2020-10-20
regression model, or type of the building energy regression model is
calculated and derived,
and reliability analysis is carried out on at least any one of R2, AdjR2, RMSE
or CV RMSE.
With regard to the amount of energy saving, use conditions of the first and
second
outdoor-air temperatures before and after execution of energy retrofit which
is comprised of
building energy saving actions including a replacement of building heat
insulating material,
an improvement in equipment or automatic control operation method, are
adjusted to be
same, thereby, the amount of energy saving is identified based on comparison
with actual
amount of the building energy consumption.
Also, in order to achieve the above objects, a method for driving a system for
measuring and evaluating building energy performance according to the present
invention
conducts measurement and evaluation of performance regarding building energy
using
building management servers A and B, and comprises following steps: receiving,
by a
measurement variable setting module, measurement data, which includes a
measurement
history, from the building management servers A and B and setting building
energy and an
outdoor-air temperature as variables; receiving, by a measurement data
analysis module, the
variables, related to occasions before and after energy retrofit for saving
the building energy,
from the measurement variable setting module and analyzing a correlation
between the
building energy and first and second outdoor-air temperatures using an
analysis graph
showing trend of and correlation between the building energy and outdoor-air
temperature
before and after energy retrofit; plotting, by a regression model generating
module, the
variables related to the occasion before energy retrofit, which have been
output from the
measurement data analysis module, generating a building energy regression
model with a
change point of the first outdoor-air temperature, and applying the second
outdoor-air
Date Recue/Date Received 2020-10-20
temperature, related to the occasion after energy retrofit, to the building
energy regression
model, thereby generating a reference model; statistically analyzing, by a
statistical analysis
module, an energy retrofit model with regard to the building energy regression
model and
the reference model, which have been output from the regression model
generating module;
and comparing, by an energy saving analysis module, a difference between a
predicted
amount of energy consumption, which is one of data resulting from statistical
analysis by the
statistical analysis module, and an actual amount of energy consumption, which
has been
output from the building management servers A and B, thereby extracting an
amount of
energy saving.
The method further comprises a step wherein the system for measuring and
evaluating
building energy performance registers one or more monitoring and control
points
corresponding to facility generating the building energy before analyzing the
correlation,
and then collects the measurement data using any one of a work tree type
configuration
method or an individual energy metering installation method for the facility.
The method further comprises a step wherein the analysis graph according to
the
correlation uses a time series plot or an X-Y scatter plot.
The method further comprises a step wherein the building energy regression
model is
set as at least any one of an average model, a simple regression model, a
heating regression
model, a cooling regression model, a cooling and heating regression model, or
a general
regression model, according to shape of the analysis graph and a number of the
change
points searched from the analysis graph.
In the method, searching for change point of the analysis graph is to
sequentially
6
Date Recue/Date Received 2020-10-20
select a virtual change point from a minimum value to a maximum value of the
measurement data and to search a factor in which sum of residual (Mean Squared
Error,
MSE) for left side data and right side data is minimum using a least square
method with
each virtual change point as center.
The method further comprises steps wherein the statistical analysis comprises:
calculating and deriving any one of a number of the measurement data, a number
of the
change points of the building energy regression model, coordinates of the
change point of
the building energy regression model, slope of the left side model or right
side model of the
building energy regression model, or type of the building energy regression
model; and
conducting a reliability analysis on at least any one of R2, AdjR2, RMSE or
CV_RMSE.
The method further comprises a step wherein with regard to the amount of
energy
saving, use conditions of the first and second outdoor-air temperatures before
and after
execution of energy retrofit which is comprised of building energy saving
actions including
a replacement of building heat insulating material, an improvement in
equipment or
automatic control operation method, are adjusted to be same, thereby, the
amount of energy
saving is identified based on comparison with actual amount of the building
energy
consumption.
ADVANTAGEOUS EFFECTS
A system for measuring and evaluating building energy performance and method
for
driving same of the present invention can verify the energy saving effect when
using good
energy saving devices or materials and improve the performance through
efficient energy
management.
7
CA 3005184 2019-08-16
A system for measuring and evaluating building energy performance and method
for
driving same of the present invention relates to a method for evaluating
building energy
performance considering a change in outdoor-air temperature that has a major
influence on
building energy consumption before and after energy retrofit of building based
on
measurement data of building management server (A and/or B), and it can secure
the
reliability of the energy saving effect through the energy retrofit of the
existing building,
thereby can guarantee the performance.
In addition, when a system for measuring and evaluating building energy
performance
and method for driving same of the present invention is applied to an existing
building
management server (A and/or B) of building, it is possible to minimize an
installation cost
of a separate monitoring sensor for collecting measurement data of
before/after energy
retrofit. Even after the energy retrofit, it can quickly and objectively
analyze the energy
saving effect through continuous commissioning and provide it to users,
thereby able to
prevent imprudent equipment replacement and able to efficient building energy
management.
BRIEF DESCRIPTION OF DRAWINGS
FIG. 1 is a block diagram of a system for measuring and evaluating building
energy
performance according to one embodiment of the present invention.
FIG. 2 is a flow chart showing a method for driving a performance evaluation
of a
system for measuring and evaluating building energy performance according to
FIG. 1.
FIG. 3 is a specific operation flow chart of measurement data analysis step
(S200)
according to FIG. 2.
FIG. 4 is an operation flow chart illustrating a reference model generating
step (S300)
8
CA 3005184 2019-08-16
and statistical analysis step (S400) according to FIG. 2.
FIG. 5 is a specific operation flow chart of a method of searching for a
change point in
a reference model generating step (S300) according to FIG. 2.
FIG. 6 is an exemplary drawing illustrating a graph of heating energy change
point
model of a system for measuring and evaluating building energy performance and
method
for driving same according to one embodiment of the present invention.
DETAILED DESCRIPTION OF THE EMBODIMENTS
Hereinafter, one embodiment of the present invention will be described.
FIG. 1 is a block diagram of a system for measuring and evaluating building
energy
performance according to one embodiment of the present invention, FIG. 2 is a
flow chart
showing a method for driving a performance evaluation of a system for
measuring and
evaluating building energy performance according to FIG. 1, FIG. 3 is a
specific operation
flow chart of measurement data analysis step (S200) according to FIG. 2, FIG.
4 is an
operation flow chart illustrating a reference model generating step (S300) and
statistical
analysis step (S400) according to FIG. 2, FIG. 5 is a specific operation flow
chart of a
method of searching for a change point in a reference model generating step
(S300)
according to FIG. 2, and FIG. 6 is an exemplary drawing illustrating a graph
of heating
energy change point model of a system for measuring and evaluating building
energy
performance and method for driving same according to one embodiment of the
present
invention.
Referring to FIG. 1, a building management server according to an embodiment
of the
present invention includes a server A 110 (a building automatic system or BAS)
and/or a
9
CA 3005184 2019-08-16
server B 120 (a building energy management system or BEMS), and it is
installed separately
from a system for measuring and evaluating building energy performance 100.
The system for measuring and evaluating building energy performance 100
includes a
building energy performance evaluation unit 200 and an input/output unit 130.
The building
energy performance evaluation unit 200 includes a measurement variable setting
module
210, a measurement data analysis module 220, a regression model generating
module 230, a
statistical analysis module 240, and an energy saving analysis module 250.
An input/output unit 140 receives data which has been output from the building
management servers 110, 120, and may output or transmit data which has been
output from
the energy saving analysis module 250 to outside.
Each of the modules 210, 220, 230, 240, and 250 may include a memory to
perform a
cotresponding function. A processor (not shown) included in the building
energy
performance evaluation unit 200 may be configured in an outside of each module
or a
processor and a memory may be integrated and included in an inside of each
module.
First, an energy retrofit of building can be defined as a set of actions for
energy saving
of building including replacing low performance windows, heat insulating
materials of wall
and equipment system (chiller, boiler, pump, etc.) of existing building or
improving
automatic control operation method, etc.
In this regard, until now, operation of a building management server A 110
(building
automation system) of existing buildings has been focused on simple operation
and
monitoring, and in case of a building management server B 120 (building energy
management system), there is a limitation in energy saving management of
facility manager,
so it is insufficient to verify energy saving effect and to improve
performance through
CA 3005184 2019-08-16
efficient energy management, actually, even when using good energy saving
equipment or
materials.
In order to objectively measure and verify the energy saving effect through
energy
retrofit of building, it is important to perform the building energy
performance evaluation
considering change of outdoor-air temperature which has a major influence on
cooling
energy or heating energy consumption of building before and after energy
retrofit.
Building energy is largely divided into cooling or heating energy, which is
influenced
by outdoor-air temperature, and base energy (appliance and lighting energy)
and other
energy, which are hardly influenced by outdoor-air temperature.
Such building energy (dependent variable) can be expressed as a simple
regression
model, which is described by outdoor-air temperature (independent variable).
If relation
between energy consumption and outdoor-air temperature is explained by only
one model,
there is no change point in the outdoor-air temperature, and if it is
explained by two models,
there is one change point (represented by two models based on a specific
fiducial outdoor-
air temperature).
Therefore, setting of the change point is an important parameter that
determines the
accuracy and reliability of the building energy regression model.
A system for measuring and evaluating building energy performance 100
according to
the present invention conducts measurement and evaluation of performance
regarding
building energy using building management servers A and B 110, 120, and
comprises: a
measurement variable setting module 210 for receiving measurement data, which
includes a
measurement history from past to present, from the building management servers
A and B
110, 120 and setting building energy and an outdoor-air temperature as
variables; a
11
CA 3005184 2019-08-16
measurement data analysis module 220 for receiving the variables, related to
occasions
before and after energy retrofit for saving the building energy, from the
measurement
variable setting module 210 and analyzing a correlation between the building
energy and
first and second outdoor-air temperatures; a regression model generating
module 230 for
using (plotting) the variables related to the occasion before energy retrofit,
which have been
output from the measurement data analysis module 220, generating a building
energy
regression model with a change point of the first outdoor-air temperature, and
applying the
second outdoor-air temperature, related to the occasion after energy retrofit,
to the building
energy regression model, thereby generating a reference model; a statistical
analysis module
240 for statistically analyzing an energy retrofit model with regard to the
building energy
regression model and the reference model, which have been output from the
regression
model generating module 230; and an energy saving analysis module 250 for
comparing a
difference between a predicted amount of energy consumption, which is one of
data
resulting from statistical analysis by the statistical analysis module 240,
and an actual
amount of energy consumption, which has been output from the building
management
servers A and B 110, 120, thereby data-analyzing the amount of energy saving.
Here, the first outdoor-air temperature and the second outdoor-air temperature
generally have different data profiles but may have exceptionally the same
data profile.
The building energy performance evaluation unit 200 registers one or more
monitoring and control points corresponding to facility generating the
building energy
before analyzing the correlation, and then collects the measurement data using
any one or
more of a method of configuring a work tree type or a method of installing an
individual
energy meter for the facility.
12
CA 3005184 2019-08-16
Based on the performance evaluation range and variables according to the
energy
retrofit, measurement data, which is related to building energy and outdoor-
air temperature
before and after energy retrofit using the building management server 110
and/or 120,
and/or a separate energy monitoring system, is collected and analyzed.
The collection of the measurement data is performed by calling the monitoring
and
control point of the existing automatic control system or by registering the
additional
installed monitoring system in the monitoring and control point of performance
evaluation
and constructing the work tree, and then, it is analyzed using a graph (e.g.
Time series plot,
X-Y scatter plot) showing trend of and correlation between the building energy
(dependent
variable) and outdoor-air temperature (independent variable) before and after
energy retrofit
based on the working period of energy retrofit which is separately registered.
In this case, the energy metering that is installed separately performs
registering an
energy retrofit work including the recorded information such as a work name
and a working
period before and after the energy retrofit is performed.
Analysis of the correlation uses a time series plot or an X-Y scatter plot.
The building energy regression model is any one or more of an average model, a
simple regression model, a heating regression model, a cooling regression
model, or a
cooling and heating regression model, according to shape of the graph and a
number of the
change points according to graph analysis of measured data.
In a method for searching a change point, virtual change points are
sequentially
selected from a minimum value to a maximum value of population and a factor in
which
sum of residual (MSE) for left side data and right side data is minimum is
searched using a
least square method with each virtual change point as center.
13
CA 3005184 2019-08-16
A left side regression model and a right side regression model of explored
factor are
analyzed thereby to set final change point (an intersection point of two
regression models).
Statistical analysis computes any one or more of a number of population data,
a
number of change points of a regression model, coordinates of change point of
a regression
model, slope of left side or right side regression model, or type of a
regression model, and
reliability analysis analyzes any one or more of R2, AdjR2, RMSE or CV_RMSE.
with regard to the amount of energy saving, use conditions of outdoor-air
temperatures before and after execution of energy retrofit which is comprised
of building
energy saving actions including improvements in heat insulating material,
equipment or
automatic control operation method, are adjusted to be same, thereby, the
amount of energy
saving is identified based on actual difference before and after execution of
energy retrofit.
Referring to FIG. 2, a driving method of a system for measuring and evaluating
building energy performance 100 according to an embodiment of the present
invention uses
the system for measuring and evaluating building energy performance 100 of
FIG. 1, and it
is understood that functions and steps to perform the measuring building
energy
performance and the driving method according to the present invention are the
same in the
system of FIG. 1 and the method of FIG. 2.
A method for measuring and evaluating building energy performance of a system
for
measuring and evaluating building energy performance 100 using the building
management
server 110, 120 includes steps of setting an energy retrofit performance
evaluation range and
variables S100, collecting and analyzing measurement data before and after
energy retrofit
S200, generating a building energy regression model and reference model
considering
change point S300, conducting a statistical analysis and a reliability
analysis for building
14
CA 3005184 2019-08-16
energy regression model and reference model S400, and analyzing building
energy saving
effect before and after energy retrofit S500.
Each step may be performed by the building energy performance evaluation unit
200
or each module.
The step S100 of setting an energy retrofit performance evaluation range and
variables
receives measurement data, which includes a measurement history from past to
present,
from the building management servers A and B 110, 120 and measures energy use.
The performance evaluation range according to the energy retrofit is basically
set
considering the collection method and range of measurement data according to
the type and
scope of the energy saving factor applied to the target building.
In addition, the dependent variable of the building energy regression model
for
evaluating the energy performance before and after energy retrofit is
classified by energy
source (electricity or gas energy, etc.) or building energy use purpose
(cooling, heating,
lighting, hot water supply, equipment, etc.) and the outdoor-air temperature
is set as an
independent variable.
In the step S200 of collecting and analyzing measurement data, the variables
are set
and measurement data of the variables before and after energy retrofit for
building energy
saving is inputted, and the correlation between the building energy and
outdoor-air
temperature is analyzed.
Referring to FIG. 3, based on the performance evaluation range and variables
according to the energy retrofit, measurement data, which is related to
building energy and
outdoor-air temperature before and after energy retrofit using the BAS 110
and/or a separate
energy monitoring system, is collected and analyzed.
CA 3005184 2019-08-16
The collection of the measurement data is performed by calling the monitoring
and
control point of the existing automatic control system (S110) or by
registering the additional
installed monitoring system (S140) in the monitoring and control point of
performance
evaluation (S120) and constructing the work tree (S130), and then, it is
analyzed using a
graph (e.g. Time series plot, X-Y scatter plot) showing trend of and
correlation between the
building energy (dependent variable) and outdoor-air temperature (independent
variable)
before and after energy retrofit based on the working period of energy
retrofit which is
separately registered (S200).
In this case, the energy metering that is installed separately (S140) performs
registering an energy retrofit work including the recorded information such as
a work name
and a working period (S160) before and after the energy retrofit is performed
(S150) in the
step (S200).
For example, in order to save the electric energy of a target building during
the
summer cooling-up period, in case of replacing two existing aged turbo
chillers (turbo
freezing machines) with a high efficiency turbo chiller and evaluating the
cooling energy
(electricity) saving effect, it is necessary to check first for operating
conditions (operation
time, etc.) related to the two existing chillers and monitoring and control
points related to
power consumption of the chillers, which have been measured by existing
automatic control
(or energy management system), and register all of them as monitoring and
control points
for performance evaluation.
However, if power consumption related to one chiller is not measured, a
separate
energy meter (electricity meter) can be installed and it is registered in a
'Monitoring and
Control point of Performance evaluation'.
16
CA 3005184 2019-08-16
In addition, registered monitoring and control points for each chiller
generate a virtual
monitoring and control point by some computing operations (e.g. the monitoring
and control
points of power consumption of a chiller 1 and a chiller 2 are summed together
so it is
reconstructed into one monitoring and control point as 'a chiller power
consumption') and
thereby construct a 'work tree' of performance evaluation.
In addition, the data measured for a certain period of time before and after
energy
retrofit based on registered work (a work name and a working period, etc.)
related to the
energy retrofit replacing the chiller is indicated in graph with correlation
between the
dependent variable (a chiller power consumption) and the independent variable
(outdoor-air
temperature).
In the step S300 of generating a reference model, the correlation is analyzed
by using
the measurement data of the variables before energy retrofit, a building
energy regression
model is created in view of the change point of outdoor-air temperature, and a
reference
model is generated by applying the outdoor-air temperature after energy
retrofit to the
regression model.
In other words, a building energy regression model in view of the change point
is
created based on the measurement data before energy retrofit, and a reference
model
(baseline) is created by applying the outdoor-air temperature after the energy
retrofit to the
regression model before energy retrofit.
In the present invention, the building energy regression model is a function
(graphically representable) of the cooling/heating energy (amount of
electricity or gas
consumption) with respect to the outdoor-air temperature before the energy
retrofit, while
the reference model (baseline), which is applying the outdoor-air temperature
after energy
17
CA 3005184 2019-08-16
retrofit to the regression model before energy retrofit, can be defined as a
predictive
regression model of the cooling/heating energy (amount of electricity or gas
consumption)
with respect to the outdoor-air temperature after energy retrofit.
In other words, it is possible to calculate (evaluate) the amount of energy
savings after
retrofit, by creating a reference model after retrofit (baseline energy model)
based on a
regression model before retrofit (Baseyear energy model), and by comparing it
with amount
of energy consumption after retrofit.
A method for creating a building energy regression model (or reference model)
considering the change point of the present invention is characterized by
further including,
first, setting a type of building energy regression model and a number of
change points and
second, searching for change point of the building energy regression model.
Referring to FIG. 4, the outdoor-air temperature is set as an independent
variable in
the population, and the cooling or heating energy is set as a dependent
variable (S200).
When selecting a building energy regression model depending on a graph of
population (S393a), it is classified into an average model, a simple
regression model, a
heating regression model, a cooling regression model, a cooling/heating
regression model
and a general regression model (S393b).
The number of change points (or turning point) of the outdoor-air temperature
of the
regression model is set (S393c), and it is determined whether the number of
change points is
0 (S393d), in response, when it is not 0, it is determined whether the number
of change
points is 1 (S393e), in response, when it is 1, the one change point is
searched (S3930, and
statistical analysis (S400) and reliability analysis (S400a) for building
energy regression
model are performed.
18
CA 3005184 2019-08-16
In this instance, if the model of step S393b is the average model or the
simple
regression model, proceed to step S400, and if the number of change points is
0 in step
S393d, proceed to step S400.
In step S393e, when the number of change points is not 1, it is determined
whether the
number of change points is 2 (S393i), in response, when it is not 2, return to
step S393c, and
when it is 2, search two change points (S393j) and proceed to step S400.
Referring to FIG. 5, a process for searching a change point of a building
energy
regression model or a reference model is classified according to the number of
change points
of regression model which is set.
First, in case of an average model and a simple regression model, which are
having a
zero (0) change point, it immediately performs statistical analysis of the
regression model
(or the average model) without going through a change point search process,
separately.
Second, in case of a regression model having one (1) change point, virtual
change
points are sequentially selected from a minimum outdoor-air temperature
(second data) to a
maximum outdoor-air temperature ((n-2)th data) of population and a factor
(outdoor-air
temperature) in which sum of residual (MSE) for left side data and right side
data is
minimum is searched using a least square method with each virtual change point
as center.
And a left side regression model and a right side regression model of explored
factor are
analyzed thereby to set final change point (an intersection point of two
regression models).
Third, in case of a regression model having two (2) change points, the process
further
includes a step of setting a virtual equilibrium point (outdoor-air
temperature) for
distinguishing the two change points, and the subsequent method for searching
a change
point is the same as the process for searching one (1) change point.
19
CA 3005184 2019-08-16
Referring to FIG. 5, a method for searching the change point sets the number
of
change points (S340) first, and then when it is 1, the one change point is
searched (S370),
and the process starts with `i (data) = 2' (S371).
It is determined whether < n' (S372), if yes, perform data setting to left
[x(1),---
,x(i)][y(1),---,y(i)] and right [x(i+1),---,x(n)][y(i+1),---,y(n)] (S373).
In the regression model, sum of MSE of left data and right data is calculated
using the
least square method `MSE(i-1)=MSE(L)+MSE(R)' (S374), then if `i = i + 1'
(S375), search
a factor i at which the MES(i) becomes a minimum *K=arg mm MSE(i)' (S376),
perform a
data setting to left [x(1),---,x(k)] [y(1),---,y(k)] and right [x(k+1),---
,x(n)][y(k+1),---,y(n)]
(S377), calculate intercepts and slopes of left regression model and right
regression model
(S378), determine a change point by performing analysis on left and right
regression model
(S379), and perform statistical analysis of the regression model (S379a).
In this instance, if the number of the change points of step S340 is 0,
proceed to step
S379a, and if the number of change points is 2, search the two change points
(S390), set a
virtual equilibrium point (outdoor-air temperature) for distinguishing the two
change points
(S390a), perform a data setting to left change point [x(1),---,x(m-1)][y(1),---
,y(m-1)] and
right change point [x(m+1),---,x(n)][y(m+1),---,y(n)] (S390b), and return to
0.
In the step S372, if it is not `i < n', proceed to step S376.
In the step S400 of conducting a statistical analysis and a reliability
analysis, the
building energy regression model and reference model are analyzed.
The number of population data, the number of change points of a regression
model,
coordinates of change point of a regression model, slope of left side or right
side regression
model, or type of a regression model, and so on are derived through
statistical analysis of the
CA 3005184 2019-08-16
building energy regression model (or reference model) including searched
change point, and
R2, AdjR2, RMSE, CVRMSE, etc. are analyzed to verify the reliability of the
regression
model.
FIG. 6 and Table 1 show examples of heating energy regression model and
statistical
analysis of the present invention.
Table 1
Left side model Result value of Right side model Result value of
statistical analysis statistical analysis
49.0000 N 51
y-intercept 20.1249 y-intercept 8.5862
Slope -2.0195 Slope 0
RMSE 0.9326 RMSE 0.9589
R2 0.9905 R2 0
Adj R2 0.9903 Adj R2 0
CVRMSE 4.1246 CVRMSE 11.1675
Change point (CP) 5.7137
The step S250 of analyzing the energy saving amount performs a statistical
analysis
and a reliability analysis, and the energy saving effect before and after
energy retrofit is
quantitatively analyzed by comparing the difference between the predicted
amount of energy
consumption of the reference model and an actual amount of energy consumption.
A building energy reference model is created by applying the outdoor-air
temperature
21
CA 3005184 2019-08-16
after energy retrofit to the created building energy regression model before
energy retrofit,
and the difference between it and an actual measured amount of energy
consumption after
energy retrofit is calculated to analyze the energy saving effect (energy
saving amount and
energy saving rate).
22
CA 3005184 2019-08-16
